Feature Spotlight

Claim Decomposition.

Turn dense articles, transcripts, and reports into a structured map of verifiable claims before you spend analyst time proving or disproving them.

What this feature does

Claim Decomposition parses a piece of content into atomic units of meaning. Instead of treating an article or statement as one block, it isolates each checkable assertion, labels it by type, and preserves the surrounding context. That gives editorial, legal, and research teams a cleaner handoff into evidence review.

Built for triage

Best used when your team needs to sort dozens of assertions quickly and decide what deserves deeper human review.

Reads Inputs

  • Breaks long narratives into discrete factual units
  • Separates claims, citations, opinions, and rhetorical framing
  • Flags unsupported leaps between evidence and conclusion
  • Builds a review queue for high-risk assertions

Produces Outputs

  • Structured claim map
  • Evidence coverage report
  • Escalation markers for legal or editorial review
  • Machine-readable JSON for downstream workflows

How to interpret results

A larger claim count does not automatically mean higher risk. The more important signal is claim density without supporting evidence, or a large number of claims grouped around one weak source.

What it cannot guarantee

Decomposition can expose ambiguous or unsupported statements, but it cannot independently prove truth. Human reviewers still need to assess source quality, context, and intent.

Start with structure before fact-checking.

This feature is designed to reduce analyst fatigue by making the review queue smaller, sharper, and easier to defend.

Request Demo